Systematic Reviews of the Clinical Effectiveness and Cost-effectiveness of Proton Pump Inhibitors in Acute Upper Gastrointestinal Bleeding

GI Leontiadis,^1† A Sreedharan,² S Dorward,³ P Barton,⁴ B Delaney,⁵ CW Howden,⁶ M Orhewere,⁷⁸VK Sharma,⁹ A Rostom,¹⁰ P Moayyedi¹¹ and D Forman^12* J Gisbert,

¹ Department of Gastroenterology, Queen Elizabeth Hospital, Tyne and Wear, UK
² Department of Gastroenterology, Lincoln County Hospital, Lincoln, UK
³ Centre for Epidemiology and Biostatistics, School of Medicine, University of Leeds, UK
⁴ Health Economics Facility, University of Birmingham, UK
⁵ Department of Primary Care and General Practice, University of Birmingham, UK
⁶ Division of Gastroenterology, Northwestern University, Chicago, Illinois, USA
⁷ Public Health Directorate, South Birmingham Primary Care Trust (NHS), UK
⁸ Gastroenterology Unit, La Princesa University Hospital, Madrid, Spain
⁹ Division of Gastroenterology and Hepatology, Mayo Clinic, Scottsdale/Phoenix, Arizona, USA
¹⁰ Division of Gastroenterology, University of Calgary, Canada
¹¹ Department of Medicine, Division of Gastroenterology, McMaster University Medical Centre, Hamilton, Canada
¹² Centre for Epidemiology and Biostatistics, School of Medicine, University of Leeds, UK

* Corresponding author
^† Present address: Division of Gastroenterology, 2nd Department of Internal Medicine, Democritus University of Thrace, Alexandroupolis, Greece

Background

Upper gastrointestinal (UGI) haemorrhage is a major cause of morbidity, mortality and medical care costs, with peptic ulcer (PU) being the most frequent source of bleeding.

It has been estimated that approximately 2–3% of duodenal ulcer (DU) patients who are not receiving antisecretory therapy are likely to develop haemorrhage during each year of follow-up, giving a cumulative risk of haemorrhage after 5 years of approximately 10–14%. Furthermore, patients whose ulcers have bled once have an increased risk of further bleeding, compared with those with uncomplicated ulcer disease. Thus, among patients who present with a bleeding ulcer, approximately one-third will develop recurrent bleeding in the following 1–2 years, and 40–50% within the subsequent 10 years, if left untreated after initial healing.

Objectives

The objectives were as follows:

• to evaluate the clinical effectiveness and cost-effectiveness of proton pump inhibitors (PPIs) in the prevention and treatment of acute UGI haemorrhage
• to evaluate the clinical effectiveness and cost-effectiveness of PPI therapy, compared with H₂-receptor antagonist (H₂RA), Helicobacter pylori eradication (in infected patients) or no therapy, for the prevention of first and/or subsequent bleeds among patients who continue to use non-steroidal anti-inflammatory drugs (NSAIDs)
• to evaluate the clinical effectiveness of PPI therapy, compared with H₂RA, H. pylori eradication (in infected patients) or no therapy, for the prevention of subsequent bleeds in patients who had previously experienced peptic ulcer bleeding.

Methods

For the first objective, evidence was sought with the Cochrane Collaboration methodology from two systematic literature reviews of randomised controlled trials (RCTs) on the clinical effectiveness of oral or intravenous PPI treatment in patients with acute UGI bleeding. The first review concerned PPI use in patients with an endoscopic diagnosis of PU; trials were identified by searching CENTRAL, the Cochrane Library, MEDLINE, EMBASE and major conference proceedings up to November 2004. The second review concerned oral or intravenous PPI use prior to endoscopy in patients with UGI bleeding; trials were identified by searching CENTRAL, the Cochrane Library, MEDLINE, EMBASE, CINAHL and major conference proceedings up to September 2005. Subsequently, data from these systematic reviews and other sources, including a patient quality of life survey in Birmingham and Leeds using the EuroQoL questionnaire, were entered into an economic model to compare the costs and quality-adjusted life-days of alternative management strategies over a 28-day period for patients who have had UGI bleeding.

For the second objective, results were obtained from two further updated Cochrane systematic reviews to (a) compare the prophylactic effectiveness (including, as an outcome, UGI haemorrhage) of PPIs, H₂RAs and prostaglandin analogues in patients requiring chronic use of NSAIDs (trials were identified by searching MEDLINE to June 2002, Current Contents for 6 months prior to August 2004, EMBASE to August 2004 and CCTR to 2004; studies on cyclooxygenase-2 selective NSAIDs were not included) and (b) compare long-term rates of first or subsequent bleeding from PU in H. pylori-positive NSAID users with either eradication therapy alone or long-term prophylactic antisecretory therapy; trials were identified by searching the Cochrane Library, MEDLINE and EMBASE up to February 2006.

Data from these systematic reviews and other sources were then entered into a Markov model with a Monte Carlo simulation to identify the most cost-effective treatment strategy for the prevention of UGI bleeding (first and subsequent) among NSAID users using an outcome of costs per quality-adjusted life-years (QALYs) over a lifetime from age 50 years.

For the third objective, results were obtained from one further updated Cochrane systematic review comparing long-term re-bleeding rates from PU in H. pylori-positive patients with either eradication therapy alone or ulcer healing antisecretory therapy followed by long-term prophylactic antisecretory therapy. Trials were identified by searching the Cochrane Library, MEDLINE, EMBASE, CINAHL and major conference proceedings up to January 2005.

Results

PPI treatment initiated after endoscopic diagnosis of PU bleeding significantly reduced re-bleeding [odds ratio (OR) 0.49, 95% confidence interval (CI) 0.37 to 0.65; number-needed-to-treat (NNT) 13, 95% CI 9 to 25] and surgery (OR 0.61, 95% CI 0.48 to 0.78; NNT 33, 95% CI 20 to 50) compared with placebo or H₂RA. There was no evidence of an overall effect of PPI treatment on all-cause mortality (OR 1.01, 95% CI 0.74 to 1.40). However, PPIs significantly reduced mortality in subgroups when studies conducted in Asia were examined in isolation (OR 0.35, 95% CI 0.16 to 0.74) or when the analysis was confined to patients with high-risk endoscopic findings (i.e. spurting bleeding, oozing of blood or a non-bleeding visible vessel) (OR 0.53, 95% CI 0.31 to 0.91).

PPI treatment initiated prior to endoscopy in UGI bleeding significantly reduced the proportion of patients with stigmata of recent haemorrhage (SRH) at index endoscopy compared with placebo or H₂RA (OR 0.67, 95% CI 0.54 to 0.84; NNT 11, 95% CI 7 to 25). However, there was no evidence that PPI treatment affected clinically important outcomes, namely mortality, re-bleeding or need for surgery.

The strategy of giving oral PPI both before and after endoscopy, with endoscopic haemostatic therapy (EHT) for those with major SRH, is likely to be the most cost-effective. Under base-case assumptions, this strategy is preferred to all other modelled strategies on cost-effectiveness grounds at any threshold over £25,000 per QALY, even if only short-term effects are taken into account, and at any threshold over £200 per life-year gained if long-term effects are included.

The risk of NSAID-induced endoscopic gastric and duodenal ulcers was reduced by standard doses of PPI [relative risk (RR) 0.40, 95% CI 0.32 to 0.51 and RR 0.19, 95% CI 0.09 to 0.37, respectively] and misoprostol (RR 0.26, 95% CI 0.17 to 0.39 and RR 0.47, 95% CI 0.33 to 0.69, respectively) and double doses of H₂RAs (RR 0.44, 95% CI 0.26 to 0.74 and RR 0.26, 95% CI 0.11 to 0.65, respectively). Standard doses of H₂RAs reduced the risk of endoscopic duodenal ulcers (RR 0.36, 95% CI 0.18 to 0.74) but not gastric ulcers (RR 0.73, 95% CI 0.50 to 1.09). PPIs reduced NSAID-induced dyspepsia. PPIs were superior to misoprostol in preventing recurrence of NSAID-induced endoscopic duodenal ulcers, but PPIs were comparable to misoprostol in preventing the recurrence of NSAID-induced endoscopic gastric ulcers. Full-dose misoprostol reduced bleeding, perforation or gastric outlet obstruction due to NSAID-induced ulcers (OR 0.60, 95% CI 0.36 to 0.98). However, misoprostol was poorly tolerated and associated with frequent adverse effects. PPIs and H₂RAs were well tolerated. However, further data comparing these agents directly with misoprostol are required to compare clinical outcomes including ulcer-related complications and prevention of NSAID-induced dyspepsia.

H. pylori eradication treatment was equally effective with PPI treatment for the primary or secondary prevention of endoscopic ulcers in NSAID users. H. pylori eradication treatment was more effective than placebo for the primary prevention of endoscopic PU and for the prevention of re-bleeding from PU in NSAID users.

With regard to primary and secondary prevention of bleeding PU in NSAID users, the two most cost-effective strategies are H. pylori eradication alone, and H. pylori eradication followed by misoprostol (substituted for a PPI, if misoprostol is not tolerated) at an additional £4810 per QALY. In a Monte Carlo analysis, H. pylori eradication strategy appears to be cost-effective compared with doing nothing for almost all patients at any level above £80/QALY. The H. pylori eradication followed by misoprostol strategy is 90% likely to be cost-effective compared with doing nothing for a willingness to pay of £500/QALY, rising to 99% at £1000/QALY. Under most reasonable assumptions, strategies involving PPI use (either alone or in combination with H. pylori eradication) are not cost-effective.

In patients who had previously experienced a bleed from a PU, re-bleeding was less frequent after H. pylori eradication therapy than after non-eradication antisecretory therapy, whether or not the latter was combined with long-term maintenance antisecretory therapy (OR 0.17, 95% CI 0.10 to 0.32 and OR 0.25, 95% CI 0.08 to 0.76, respectively). The NNT to prevent one episode of re-bleeding with eradication therapy was 7 (95% CI 5 to 11) compared with ulcer healing treatment alone, and 20 (95% CI 12 to 100) when compared with long-term maintenance antisecretory therapy.

Conclusions

PPI treatment compared with placebo or H₂RA reduces mortality following PU bleeding among patients with high-risk endoscopic findings. It also consistently reduces rates of re-bleeding and the need for surgical intervention. PPI treatment initiated prior to endoscopy in UGI bleeding significantly reduces the proportion of patients with SRH at index endoscopy but does not reduce mortality, re-bleeding or the need for surgery. The strategy of giving oral PPI before and after endoscopy, with EHT for those with major SRH, is likely to be the most cost-effective.

Treatment of H. pylori infection is more effective than antisecretory therapy (with or without long-term maintenance antisecretory therapy) in preventing recurrent bleeding from PU. H. pylori eradication alone or H. pylori eradication followed by misoprostol (with switch to PPI, if misoprostol is not tolerated) are the two most cost-effective strategies for preventing bleeding ulcers among H. pylori-infected NSAID users, although the data cannot exclude PPIs also being cost-effective.